Customer Stories: Investment AI firm uses machine learning and Microsoft startup support to drive big business at low cost
This post was created by the Microsoft Customer Stories team and was originally posted on customers.microsoft.com
“We’re looking at about $500,000 in revenue, made possible by Microsoft working with us. And that’s just one client for one year…. Thanks to Azure and the Microsoft for Startups program, everybody really does win.”
- Leslie Kanthan: Chief Executive Officer
Turing Intelligence Technology (TurinTech) is an artificial intelligence company that provides an algorithmic trading platform offering automation based on conditions and logic rules, signal, trade, and event-driven trade execution, transparent trades, auto custom exchange trade, and trading that relies on complex machine learning models. Or, without the jargon: our company provides an easy way for traders to develop and implement ideas about when and how to buy and sell securities. TurinTech opens advanced trading to personal investors as well as institutional investors.
The need for an open trading platform
Three years ago, the founders of TurinTech recognized a need for an open trading platform. Historically, financial institutions have used either an in-house proprietary trading system or one from an external vendor. Both are closed, with few people to maintain them. If an expert in a particular trading system goes to work for another bank, that person has to learn a whole new system because algorithms and coding are specific to each organization. And this isn’t just about big investment banks. Many midsize firms, family offices, trusts, small hedge funds, and brokers would benefit from having market choices.
For example, suppose a trader wants to take into account hot topics in social media, the current popular search terms on Google, the latest unemployment figures—whatever the source might be—in real time. TurinTech provides access to many such data streams in addition to up-to-date equity pricing, corporate facts and figures, and futures, to name a few. We collect and collate the information using machine learning and offer simple drag-and-drop tools and APIs to build an organization’s solutions on top of our platform. Plus, we provide free back-testing (which examines historical data to assess risk) in real time.
Trust and convenience from blockchain
To ensure trust, TurinTech users can code and test algorithms on our platform without anyone (including us) seeing them. We provide cryptographic proof of that fact. When a trade happens, an immutable and irreversible transaction is stored in a blockchain. Of course, the TurinTech platform can connect to traditional brokers’ back-end systems and perform trades as usual, but the interesting part of our platform is its ability to use a blockchain network and resolve trade disputes quickly. Because the TurinTech platform is built on Microsoft Azure, using it reduces an organization’s IT labor resources and costs. Also, our platform handles back-office functions—settling trades, verifying payments, and checking for trading irregularities. It’s self-perpetuating and self-maintaining.
The right technology for intelligent algorithms
The first machine learning algorithms TurinTech used came from Microsoft Cognitive Services, which outperformed similar technologies that we tried. We still use it to identify specific keywords and relevant information sources. We’ve also developed our own machine learning algorithms to help users extract and put to work keyword-associated text, whether it’s from web pages, PDF files, or other documents.
The solid performance of blockchain
A big issue for many blockchain systems is speed. TurinTech will soon have an enterprise-ready blockchain implementation that takes advantage of Azure virtual machines, which we believe are the fastest, most reliable on the market. We’re already demonstrating high-performance blockchain proofs of concept to key customers.
We’re especially proud of the ability of the TurinTech platform to automatically downsize customers’ code, using machine learning to automate refactoring and deduplication. For example, suppose an analyst for a bank creates code that performs a statistical analysis on a set of data. Another analyst in another location writes similar code, not knowing that someone else has already done it. Later, when the two code blocks are invoked, TurinTech will be able to identify and remove duplicate programming tasks and downsize code by up to 50 percent. And if we’re looking at, say, 10 million lines, that’s a huge factor in reducing subscription costs. This capability is a work in progress that we expect to deliver to market in about three months, but we’re already testing it in multiple organizations and it’s generating revenue.
A relationship that profits all stakeholders
Even before Microsoft introduced us to its Microsoft for Startups program, we had a close relationship with the company’s research division. We had no idea that the startup program would connect us to Azure sales reps who were already working directly with the institutions we hoped to reach. That was fantastic. Microsoft gave us the chance to sit down with major financial players and tell them about our projects. We now have a symbiotic relationship with Microsoft. Azure reps are already connected to our potential customers, and Microsoft is interested in selling our products along with Azure solutions. That means Microsoft will visit some clients on our behalf, so we won’t have to give the initial sales pitches.
To put numbers on that, in one case Microsoft approached one of its big bank customers and explained how running TurinTech on top of Azure is a great value. Our initial outlay was zero because Microsoft is already there. When a final agreement is in place, we’re looking at about $500,000 in revenue, made possible by Microsoft working with us. And that’s just one client for one year. The bank gets a great deal, we get a new customer, and Microsoft delivers a cutting-edge solution on Azure. Thanks to Azure and the Microsoft for Startups program, everybody really does win.
Plans for future growth
As we make plans to work with more global firms, the next step for TurinTech is to enter the US and China markets. Microsoft will be able to introduce our company to counterpart clients in those countries, thanks to its existing strong presence there. Penetrating the China market alone, for example, will allow TurinTech to open business channels into more than 30,000 hedge funds. Many of these funds are connected to high-profile clients. It will also connect us to an individual investor population of more than 200 million people. In turn, this will exponentially and cost-effectively increase our use of Azure resources, which we value and trust given the platform’s flexibility.
“TurinTech will be able to identify and remove duplicate programming tasks and downsize code by up to 50 percent. And if we’re looking at, say, 10 million lines, that’s a huge factor in reducing subscription costs.”
- Fan Wu: Chief Scientist
- Turing Intelligence Technology